Translation Errors in Crouwdsourced Translations: an Analysis of Kiswahili Fcebook User Interface Content

Abstract

This study is set out to investigate translation errors in crowdsourced translations for Facebook’s Kiswahili user interface. These errors were identified, categorized and their apparent assessed. In order to achieve the study’s purpose, data on translation errors was collected from Facebook user interface by taking screenshots of texts that were selected through purposive sampling. Translation errors were then identified based on the basic concepts of the Dynamic Equivalence Theory proposed by Eugine Nida, and the content analysis method. The errors were grouped into three categories; grammar errors, stylistic errors and accuracy errors. The analysis of these errors revealed that they arose mainly from literal translation, lack of provision of a context on the part of Facebook and lack of translator training. The study also assessed apparent translation strategies that were used by translators which may have led to these errors. These were mainly word-for-word translation and oblique translation. Moreover, the study provided a basis for further studies by recommending the use of a hybrid translation approach between crowdsourcing and outsourcing, arriving at a stronger translation approach which leverages on the strengths of both approaches.